Genetic algorithm applied to state feedback control design

Abstract

Control design to MIMO systems requires the state space representation. In the state space, the control strategy is the states feedback and the most used techniques are the LQR and the LQG. In despite of the good results obtained from these methods, the control design is not a straightforward task due to the trial and error method involved in the definition of weight matrices. In such cases, may be hard tuning the controller parameters in order to obtain the optimal behavior of the system. In this work, it proposes a states feedback technique in which there are no trial and error processes involved and the control design is carried out to fulfill specifications, for maximum overshoot and accommodation time. The proposed technique is based on the modification of the states transition matrix and uses genetic algorithms. The obtained results show that is possible to design controllers which fulfill design specifications.